Criterios funcionales

Measuring acoustic parameters is a wide spread method to compare dolphin whistles, and it can give information about the general position of a signal in a multivariable space. However, it is important to note that parameter correlations might be high for whistles that have dissimilar contour modulation patterns. Likewise, the contour match can indicate a similar modulation pattern regardless of the actual position of the contour in frequency-time space. This is important considering that signature whistles can convey identity information in their modulation patterns independent of vocal cues (Janik et al., 2006). Similarity values calculated using parameters showed no significant values in any analysis, and showed low correlation. Acoustic parameters in signature whistles seem to vary with context (Janik et al., 1994) whilst modulation patterns are

relatively stable as shown in Chapter 4 in this thesis. King et al. (2013) showed that dolphins copy

In d e x o f S im ila ri ty ( C o n to u r) In d e x o f S im ila ri ty ( P a ra m e te rs ) 1 2 3 1 2 3 A B

Chapter 5. Influence of relatedness and calf association on signature whistle similarity

106

signature whistle contour shapes of close associates, but without matching all acoustic parameters of the copier’s whistle. Likewise, Ralston & Herman, (1995) showed that dolphins seem to be capable of processing the whistle contour without using the absolute frequency. Taking this in account, acoustic parameters do not necessarily reflect the modulation pattern of the whistle, and dissimilar contours might show absolute frequency similarity. Therefore, contour similarity using dynamic time warping algorithm stated in (Deecke & Janik, 2006) seems to be better method to represent for signature whistle similarity than using parameters correlation.

Studying signature whistle contour similarity amongst bottlenose dolphins can help to unravel how individuals learn their signature whistle and how the community influences the vocal learning process, especially taking in account that the modulation pattern is sufficient to convey individual information (Janik et al., 2006). In this study, the significant results of the Mantel test for coefficient of relationship and contour similarity (Figure 5.1) indicates that generic genetic closeness increases the probability of two individuals of having similar whistle contours. However, this significant result of a relationship of the index of relationship and whistle contour similarity could be confounded with the significant relationship of similarity found for mothers and calves. The mother and calf degree of contour similarity does not necessarily has to be based on the genetic closeness, but on a high association in early life (Wells & Scott, 1999; Mann, 2000) as I will discuss below, especially taking in account that the results of the binomial Mantel tests for siblings, grandmothers, and aunts or uncles showed that those kin relations are not more likely to have similar whistles than unrelated individuals (Figure 5.2, Figure 5.4-Figure 5.6).

Mother and calf pairs showed a significant positive correlation with contour similarity and the fact that they are the kin with the highest coefficient of relatedness could explain the result of the analysis for genetic similarity. In this case, the results showing that individuals have more

Chapter 5. Influence of relatedness and calf association on signature whistle similarity

107

probability to have similar whistles to their mothers can be explained by their degree of association in their youngest years. Dolphin calves of both sexes spend up to 5 years of their lives highly associated with their mothers until weaning, while slowly the calves become more

independent and start associating with other individuals (Connor et al., 2000; Wells, 2009). Sayigh

et al. (1995) found that male calves tended to have more similar whistles to their mothers than female calves, although not all the male calves of the same female necessarily developed a signature whistle similar to their mothers. However, in this study, there was no significant difference in the median similarity values between calves of both sexes neither when using contour or parameter similarity (Figure 5.3), suggesting that males and females have similar probability of developing a signature whistle similar to their mothers. Both sexes are exposed to their mother’s signature whistle during the first year of life, and the identity of the mother could

be influencing more whether a calf develops a similar or dissimilar whistle than just sex (Sayigh et

al., 1995). The discrepancy of results with Sayigh et al. (1995) could be related most likely to the difference in the method of calculating similarity between both studies and the smaller span of years sampled (1976-1990) in comparison with this thesis. Sayigh et al. (1995) calculated as a categorical rank of similarity going from 1 (not similar) to 5 (very similar) using human observers. I compared 44 calves (1984-2013) in this study using a quantitative method of similarity match. The limited qualitative scale (categorical variable) could drive observers to rank whistles with a relative low similarity for female calves, that quantitatively (using DTW), would have a higher percentage similarity match (continuous variable).

In addition, the lack of high similarity values between calves and grandmothers compared with unrelated individuals, but high similarity between mothers and calves could suggest that dolphins select sections of their mother’s signature whistle to model their own, but these sections

Chapter 5. Influence of relatedness and calf association on signature whistle similarity

108

are not necessarily the same as the ones used by the next generation (grandchildren), avoiding of a matrilineal tendency. Ultimately, a signature call is meant to be sufficiently distinctive to determine an individual (Shapiro, 2010). In the fission-fusion social structure of bottlenose dolphin populations, kin relationships other than mothers with their calves are not necessarily bonded

strongly. For example, in Sarasota, male alliances are usually between unrelated males (Connor et

al., 2000). In contrast, other dolphins like killer whales tend to maintain pod-specific repertoires (Yurk et al., 2002; Miller et al., 2004), but this is related to matriarchal defined stable groups, a situation that does not occur in bottlenose dolphins. This lack of kin bonds might suggest that the degree of association would become more relevant for influencing acoustic similarities than the genetic relatedness itself. This premise is supported by the results in this study showing no significant correlation between signature whistle similarity and any kin other than mothers. Thus, the similarity in signature whistles in bottlenose dolphins might be influenced more by the social interaction rather than maternal kinship.

Associations within bottlenose dolphin communities can vary greatly across the years (Irvine et al., 1981; Wells et al., 1987; Bräger et al., 1994). Sometimes, the limitation of comparative data across years when studying associations might lead to non-representative results. For instance, the lack of significance with strong correlation values for the results of Mantel tests when comparing index of association and whistle similarity by year (Figures 5.7), could be explained by the short range in the index of association values of specific individuals sampled. For example, the Mantel test results for the individuals in the year 2000 (Figure 5.7 A) and 2010 (5.7 B) showed not relationship between index of association and whistle similarity. However, the values for index of association ranged only from 0.0 to 0.18, a very low value in a narrow range. In contrast, the model that fit with the highest correlation and the lowest p value is

Chapter 5. Influence of relatedness and calf association on signature whistle similarity

109

the one for year 1999 (Figure 5.7 C). It showed the broadest range for index of association values of all years (0.0 to 0.45). In addition, for analysis of independent years the sample size might also have an effect in calculating the yearly relationship accurately. Therefore, the relationship of signature whistle similarity and association becomes clearer with the when all the pairs from each year were combined in a Spearman’s Correlation test (Figure 4.8) and the range of coefficients of association ranged from 0.0 to 0.55.

The values used for the association analyses were only representative for the first year of life, which is considered the moment when dolphins are developing their signature whistle modulation pattern (Killebrew et al., 2001; Fripp & Tyack, 2008). An influence of those associations on signature whistle modulation patterns might be seen during the youngest period of a dolphin’s life. Bottlenose dolphins are vocal learners with the capacity of innovate sounds to add to their repertoires (Janik, 2014). Therefore, associations with individuals during the first year of life might have an important role in shaping signature whistles, especially when the individuals associated are also one year old. Calves associate with all the individuals that their mother associates with and the sounds those individuals produce appears to influence the repertoire of the calves (Fripp

et al., 2005). Many juveniles in Sarasota show a high philopatry and once they are independent from their mothers, they tend to associate with individuals that they used to interact with when they were calves (McHugh et al., 2011). My results suggest that these continuous associations seem to play some role in the development of signature whistle modulation patterns. It has been shown that individuals in captivity spontaneously use acoustic signals from their surroundings to add new signals to their repertoires (Tyack & Sayigh, 1997; Miksis et al., 2002). Also, social companions have a strong influence on whistle similarity in whistle repertoires in captive adult

Chapter 5. Influence of relatedness and calf association on signature whistle similarity

110

found that calves in Sarasota show a higher whistle contour similarity with individuals which associate regularly but not the most in the community. Their results suggest that individuals that are encountered a lot might not be the main ones influencing signature whistle modulation in calves. However, Fripp et al. (2005) investigated all associates of the calves for one single year, and the measurement of association was established as percentage of time. For this study, I used coefficients of association (CoA) based in several encounters, and compared a larger number of calves. In addition, I only compared newborn calves with one another. There seems to be an influence of calves on calves, that is not apparent amongst adults and calves.

I also found that the sex of calves is not relevant for the degree of whistle similarity displayed by a pair, as pairs consisting of different or same sex calves showed no significant difference in their contour or parameter similarity (Kruskal-Wallis Test, p>0.05 ) (Figure 5.9). Therefore, the association between calves itself, regardless of sex of the other individual, is what might have influence on the similarity of signature whistle contours amongst one year old calves. However, this might change when males become independent and reach sexual maturity, as male alliances seem to modify certain characteristics in their whistle contour to converge whistle repertoires (Smolker & Pepper, 1999; Watwood et al., 2004).

In conclusion, using the contour comparison with dynamic time warping seems to be a better method to calculate similarity between signature whistles than using parameter pairwise correlations, especially when the interest is about modulation patterns. In addition, this study showed that the high coefficients of associations amongst unrelated calves correlate significantly with high signature whistle contour similarity. On the other hand, mother and calf pairs are more likely to show high whistle similarity, but the same is not true for other kin relationships (sibling, grandmothers and calves, and uncles or aunts and calves) which did not show significant

Chapter 5. Influence of relatedness and calf association on signature whistle similarity

111

correlation between coefficient of relatedness and contour similarity. Therefore, considering that mothers are the highest associated individual of their offspring during the first year of life (Mann, 2000), it could be suggested that social interaction between individuals in highly dynamic societies, especially at very young age, seems to have more influence on signature whistle crystallisation and modulation changes than relatedness itself.

Chapter 6. Effects of carrying a tag on bottlenose dolphin signature whistles and surface behaviour

112

CHAPTER 6

EFFECTS OF CARRYING A TAG ON BOTTLENOSE DOLPHIN SIGNATURE WHISTLES AND